Principal findings
Our results demonstrate that people with probable depression in Central Nepal used more healthcare and incurred greater OOP expenditures than people without depression, and that both utilization and OOP expenditure increased significantly with increasing depression screening scores. People with probable depression incurred total mean expenditures of $118 (SD 29) USD/year OOP on healthcare, which is likely to represent a substantial proportion of household budgets considering that median annual income for adults with depression in this area has been estimated to be $501.17 The association between symptom severity and total healthcare utilization and OOP costs was attributable to outpatient rather than inpatient service use, and was not modified by relative wealth.
Our findings also show that when seeking healthcare, individuals with probable depression most often sought care from pharmacists, with very few consulting mental health specialists for healthcare. Despite their more frequent use, annual pharmacist costs were much lower than for consultations with specialist doctors ($7.97 USD/year compared to $36.99 USD/year) and general doctors ($12.04 USD/year).
Strengths and limitations
This is the only population-based study that we are aware of from Nepal or any other low-income country to simultaneously evaluate dose-dependent associations between depression and healthcare utilization and OOP expenditure and to examine sector-specific utilization. We used a large, population-based sample, in which secondary level educational attainment was fairly representative of national estimates, suggesting that this was not an obvious source of selection bias.”46We also minimised the risk of misclassification bias because we validated the PHQ–9 in this setting and modelled symptom score as a continuous variable for most analyses.
However, our findings should be interpreted in light of some limitations. Firstly, due to the cross-sectional design, (and because the recall periods for the PHQ–9 and healthcare utilization were 2 months and 3 to 12 months respectively), it is difficult to infer causality or determine the true temporal relationship between depression and healthcare utilization. Multiple testing may also have led to spurious results, although the pattern of results was consistent across outcomes and similar to other studies from other resource-poor settings.22,37,47 However, depression typically adopts a chronic relapsing and remitting course and it is likely that these symptoms preceded the 2 week recall period.” . Secondly, in the absence of reliable health service records, we measured utilization and expenditure by self-report, which may be subject to recall bias, although we restricted the recall period for outpatient utilization to three months to minimise this. Conversely, by extrapolating outpatient healthcare utilization from 3 months to 1 year, we may have introduced infrequency bias due to seasonal fluctuations, which could have led us to over or underestimate these associations. Thirdly, we cannot rule out unmeasured confounding from urban or rural residence or physical comorbidities, which are known to influence healthcare access in Nepal48,49, as neither were recorded in this study. Consequently, we are unable to quantify the relative contributions of other serious health conditions to healthcare utilization and OOP expenditure. However, previous studies that have adjusted for comorbidities have nevertheless found residual associations22, while studies restricted to populations with chronic conditions still find increased healthcare expenditures among people with mental illness50. Finally, the OOP expenditures that we report are derived from healthcare utilization by the interviewee alone but may impose a wider burden to the household and to friends, whose associated costs were also included. However, we did not collect income or expenditure data at the household level and were therefore unable to model either expenditures as a proportion of available household budget, or catastrophic healthcare expenditures, which are more likely to have a direct impact on family members.
Comparison with previous literature
The pattern of increased healthcare consumption by people with probable depression in Nepal reported here is consistent with observations from high-income countries, where the greatest direct healthcare costs are also attributable to general healthcare rather than specialist mental healthcare utilization.12 Our finding that excess expenditure by people with depression was largely accounted for by outpatient service use is consistent with findings from India21,51, Brazil52 and other middle-income countries22, which demonstrate 14 to 36% increases in community healthcare utilization among people with depression, and suggest that depression is a risk factor for catastrophic health expenditure.37 This is the first study that we know of to independently report hospital admission frequency and OOP healthcare expenditures in people with symptoms of depression in LMIC, although a study from Canada also found no significant increases in health service related costs of admissions.12
Whether people with depression seek more help from pharmacists in LMIC has not been studied before, but in Brazil it has been reported that depressed individuals use more medications than the general population.52 It is important to note that the role of pharmacists in Nepal is generally limited to the provision of over the counter drugs rather than diagnoses or prescriptions. As reported elsewhere in Nepal48, we detected minimal utilization of traditional healers for overall healthcare needs. This may be due to increased stigmatization of the use of traditional healers53 and a general preference for ‘pill-based’ treatments in South Asia54. Conversely, only 2% of people with depression who utilized outpatient services were seen by a psychiatrist, which is consistent with previous literature and the shortage of mental health specialists in the area32.
Mechanisms and implications
Given the limitations of the data described above, there are multiple potential explanations of the association between depression symptoms and healthcare utilization and expenditure, which are not necessarily mutually exclusive: (1) The first explanation is that depression leads to increased utilization of services through somatization—that is, the manifestation of psychological distress as physical symptoms—which has been demonstrated previously55 (figure 3a). (2) Alternatively, depression may lead to increased healthcare utilization because of its association with physical comorbidities, which are both a risk factor for and a consequence of depression56–58 (figure 3b). (3) Finally, OOP healthcare expenditure can result in poverty38, which in turn is a major risk factor for depression16, raising the possibility that the causal pathway acts in the opposite direction (figure 3c).
There is evidence to support the plausibility of each of these pathways. In support of the first and second hypotheses, which posit a causal role of depression, associations have been shown prospectively between depression and health service utilization in Canada12 and the United States.59 Further, somatization is a common presentation of depression in South Asia60–64 and particularly in Nepal65, which could explain the increased demand for physical rather than mental healthcare. Previous analyses have also shown that only 8.1% of people with probable depression in this sample reported actively seeking treatment specifically for depression symptoms66, which is far lower than the 71% who sought any healthcare at all. This may relate to the lack of effective detection and treatment of depression in primary care67,68, and possibly to low perceived need for intervention for psychological symptoms.69
At the same time, a growing evidence base also demonstrates that depression increases the risk of a variety of physical health problems70–73, which could mediate the relationship between depression and healthcare utilization, although we were unable to explore this. Cross-sectional evidence from Nepal supports an association between depression and chronic comorbidities74,75. Nonetheless, studies that have controlled for these comorbidities or restricted to populations with chronic conditions have still found a residual association between depression and increased healthcare utilization, suggesting that depression also exerts an independent effect on treatment-seeking.22
Recent evidence shows that primary care workers can be trained24 to improve detection of depression and deliver mhGAP-based interventions26 (including psychoeducation and medication) to effectively treat depression and reduce disability in Nepal25. If depression is the primary driver of increased healthcare utilization among this group (either directly, by somatization, or indirectly, by impacting on general health) then we would expect the roll-out of such services to reduce excess healthcare use and associated OOP expenditure. Elsewhere, the integration of mental health services into primary care in Andhra Pradesh in India39 has been shown to substantially reduce OOP healthcare expenditures. In Nepal, the estimated combined costs of scaling-up services for psychosis, depression and epilepsy are less than $1.30 USD/capita/year.27 In contrast to the additional $9 USD/year OOP incurred by people in this study for each increment in PHQ–9 score, this is a relatively small investment, which might minimise financial impoverishment, whilst reducing symptoms of depression and hopefully break the link between the two.
However, the notion that OOP healthcare expenditures lead to depression (figure 3c) also finds some empirical support. Healthcare expenditure represents an exceptionally large share of total household expenditure in Nepal15 and therefore poses a risk of catastrophic expenditure, impoverishment and widened inequalities15, which are significant determinants of depression.16 In support of this, we found that the poorest individuals were more likely to suffer from depression but did not appear to forego care, which we would expect to cause further impoverishment. Thus, the lack of financial protection for people with depression is likely to reinforce a vicious cycle of further healthcare utilization and greater impoverishment. According to this model, introducing financial protections would reduce depression prevalence. Supporting the uptake of social health insurance schemes in Nepal could therefore break the cycle between the impoverishing effects of healthcare utilization and the associated risks of depression. In support of this hypothesis, studies from the United States show that introduction of Medicaid led to 9% reductions in depression prevalence and near elimination of catastrophic expenditure.76
Given previous evidence of the inter-relationships between poverty and financial shocks, depression, and physical health, it seems probable that the observed association between depression and healthcare utilization and expenditure results from a combination of these pathways (figure 3d). Indeed, the idea that physical health, mental health and poverty interact synergistically through multiple complex pathways underlies syndemics theory.77–79 As we did not collect data on all of these factors, further research would be necessary to empirically confirm the relative contribution of each of these mechanisms in the Nepali context. However, a combination of increased financial protection for healthcare and the provision of effective depression treatment might reasonably be recommended to break this cycle.
In terms of how to provide depression treatment as part of UHC, the low frequency of hospital admissions support proposals to integrate mental health services into primary rather than secondary care in LMIC. The low cost and frequent consultation of pharmacists, relative to general and specialist doctors, is also relevant to policy as it indicates an additional cadre of healthcare providers who are in frequent contact with the target population. Evidence from PRIME demonstrates that task-sharing by training and supervising non-specialist health workers can improve depression detection rates68 and mhGAP-based interventions for depression in Nepal have been found to be most effective when they were supplemented with psychosocial treatments delivered in the community.24 Our findings suggest that enlisting pharmacists to identify potential depression patients and refer them to community counsellors to receive appropriate treatments such as the Healthy Activity Program80, might be a fruitful strategy to both reduce the provision of unnecessary medications (and therefore OOP costs) and ultimately provide both mental and physical health benefits.25